The study population consisted of elderly individuals residing in a nursing home between December 2024 and August 2025. The sample size of 30 participants was determined by recruiting individuals who were super-elderly, had relatively high care needs, and were in the older age group, and by referencing previous studies investigating tongue cleaning methods (typically 20 - 40 participants). In addition, because each participant underwent all six cleaning conditions (within-subject design), the statistical efficiency for detecting differences in microbleeding incidence between devices and stroke numbers was increased.

Exclusion criteria included the presence of mucosal diseases such as oral lichen planus or oral candidiasis, history of Sjögren’s syndrome or head and neck radiotherapy that could cause oral dryness, inability to cooperate with oral hygiene management due to dementia or other conditions, and current smoker.

DENT.EX Systema Genki f (Lion Dental Products Co., Tokyo, Japan) toothbrushes were used. The bristle material was saturated polyester resin and classified as medium hardness. The tongue brush employed was a Dental Pro^® Tongue Brush (Dental Pro Co., Osaka, Japan), with nylon bristles classified as soft. As for the sponge brush, that was a Mouth Pure^® Oral Care Sponge, size S (Kawamoto Corporation, Osaka, Japan), made of urethane. Uropaper^® III H urine test strips (Eiken Chemical Co., Tokyo, Japan) were used for detection of microbleeding. Microbleeding was assessed using commercially available urine occult blood test strips. The test kit used for detecting microbleeding in the oral cavity is currently difficult to obtain; therefore, urine occult blood test strips were used as an alternative method. The urine test strips demonstrated high sensitivity for detecting trace amounts of blood, and occasionally showed a faint reaction even before tongue cleaning. Consequently, microbleeding was evaluated based on the color change of the test strips before and after cleaning.

Measurements were performed after 90 minutes of fasting, to allow for salivary pH to return to baseline [21]. Immediately after mouth opening, the examiner wiped the tongue three times using a sponge brush, with the collected material then suspended in 1 mL of water and used as the baseline specimen. A urine test strip was then immersed into the specimen and the resulting color tone used as the reference.

Next, while maintaining mouth opening, the tongue mucosa was scraped a predetermined number of times using either a toothbrush, tongue brush, or sponge brush at a site different from the area previously wiped with the sponge brush. The tongue mucosa was then wiped three times using a new sponge brush for collection, suspended in 1 mL of water, and used as the recovery specimen. A test strip was subsequently immersed in the recovery specimen, and the color tone was compared with the reference to determine the presence or absence of microbleeding. Following the manufacturer’s instructions, the test strip was read 30 seconds after sample application and evaluated using the kit’s included 5-level standard color chart. To establish individual baseline values, samples immediately after opening the mouth were tested with test papers. The color of these test papers became the baseline value for each subject. Under each experimental condition, specimens were collected and examined using the same procedure. “Positive” microbleeding was defined as an increase in grade on the color chart compared to the baseline value.

All color evaluations were conducted by a single trained inspector. Evaluations were not blinded with respect to cleaning equipment or number of strokes. To ensure consistency, standardized color charts were used for all evaluations.

For all scraping and collection procedures, wiping was performed from the posterior toward the tip. It was confirmed in advance that the series of collection procedures used with the sponge brush did not induce new microbleeding. Additionally, testing performed prior to the examinations confirmed that no color change occurred when the test strip was applied directly to the sponge brush used for wiping, while immersion of the test strip in the 1 mL water suspension resulted in a color change.

The following six experimental conditions were used: toothbrush, 3 strokes (TB3); toothbrush, 10 strokes (TB10); tongue brush, 10 strokes (TongueB10); tongue brush, 30 strokes (TongueB30); sponge brush, 10 strokes (S10); and sponge brush, 30 strokes (S30). These conditions were performed in a fixed sequence. To minimize potential carryover effects, evaluations under each condition were conducted at least two days apart [22][23]. This interval was set based on prior reports indicating that the tongue surface and tongue coating return to baseline values within several days after tongue cleaning. All evaluations were performed after confirming that the tongue coating had sufficiently reformed. Before use, each brush was moistened with water and excess moisture was removed with a paper towel.

Previous studies have noted that the mean applied pressure during tongue cleaning was 95.0 ± 40.8 g in individuals with cleaning habits and 90.9 ± 66.2 g in those without such habits [20], and that no microbleeding occurred with tongue brushing at 100 - 150 g [16]. Therefore, scraping pressure for the present participants was standardized at 100 g for all conditions. Sufficient practice was conducted to ensure consistent application of 100 g of pressure before the examined measurements and a brushing force measurement device was used (Sanei ME Co., Kanagawa, Japan). All measurements and specimen collections were performed by the same examiner. Based on previous studies suggesting that up to 30 strokes are appropriate for tongue cleaning [16], the maximum number of strokes was set at 30.

Participant medical-related characteristics were collected from medical records and included age, sex, height, weight, medical history, medications, level of nursing care required, route of nutritional intake, and serum albumin levels. Nutritional status was evaluated using the Geriatric Nutritional Risk Index (GNRI) [24]. Oral assessments included number of present teeth, clinical diagnostic criteria for oral dryness [25], oral moisture levels [26] measured using a Mucus^® oral moisture meter (Life Co., Saitama, Japan), and tongue coating index [27]. Subjective pain during tongue scraping was evaluated using the Visual Analog Scale (VAS).

Statistical analyses were performed using IBM SPSS Statistics, version 28.0 (IBM Corp., Armonk, NY, USA), with a t-test and Fisher’s exact test used as appropriate, and statistical significance defined as p < 0.05.

This study was approved by the Research Ethics Committee of Kyushu Dental University (Approval No. 22-26). Written informed consent was obtained from each participant after explanation of the study using a consent form.

The participants included 30 elderly individuals, aged 65 - 101 years, who required nursing care (6 males and 24 females; mean age 89.4 ± 7.9 years) (Table 1). As for long-term care level, no participants required Support Level 1 or 2, two required Care Level 1, three required Care Level 2, nine required Care Level 3, eight required Care Level 4, and eight required Care Level 5. The mean care level value was 3.57 (Table 2). All participants routinely received oral nutritional intake. Pain during tongue scraping was assessed and all were VAS 0, thus no participant experienced pain.

Serum albumin data were available for 27 of the participants. Nutritional status assessed using the GNRI showed no nutritional risk in five, mild risk in nine, moderate risk in seven, and severe risk in six (Table 3). Regarding medical history, hypertension was the most common cardiovascular condition (n = 27), followed by fractures/joint disorders (n = 24) and gastrointestinal disorders (n = 23) (Table 4). The mean number of prescribed medications, excluding eye drops and topical agents, was 8.5 ± 3.6 per participant, with cardiovascular medications the most frequently noted (n = 26) (Table 5).

Table 1.Participant characteristics (n = 30).

Table 2.Long-term support and care level (n = 30).

Table 3.Nutritional risk according to GNRI (n = 27).

GNRI formula: GNRI = (14.89 × serum albumin [g/dL]) + (41.7 × current body weight [kg] / ideal body weight [kg]). Classification: No nutritional risk: >98; Mild risk: 92 to <98; Moderate risk: 82 to <92; Severe risk: <82.

Table 4.Medical history (n = 30).

Fisher’s exact test. Among cardiovascular diseases, hypertension was the most prevalent, observed in 27 participants. No significant association was found between medical history and number of participants with microbleeding.

Table 5.Medications (n = 30).

The mean number of medications was 8.5 ± 3.6. Cardiovascular agents were the most frequently prescribed medications, used by 26 participants. No significant association was observed between medication use and number of participants with microbleeding.

In the TB3 group, microbleeding was observed in 12 participants (40.0%), whereas that was observed in 19 (63.3%) in TB10, a significant difference. Microbleeding was observed in eight participants (26.7%) in TongueB10 and in 18 (60.0%) in TongueB30, again a significant difference. In contrast, microbleeding was not significantly different between the S10 and S30 groups, as that was observed in two (6.7%), and four (13.3%) participants, respectively.

Comparisons of the 10-stroke groups showed that TB10 had a significantly higher number of participants with microbleeding than TongueB10 and S10. Although TongueB10 showed a tendency toward greater microbleeding than S10, the difference was not statistically significant. As for the 30-stroke groups, TongueB30 showed a significantly higher number of participants with microbleeding as compared with S30. These results are shown in Table 6.

We performed a generalized linear mixed model with a logit link function, incorporating device type, stroke count, and their interaction as fixed effects, and participant ID as a random effect. The analysis revealed significant main effects for both device type (p < 0.001) and stroke count (p < 0.001), while the interaction was not significant (p = 0.126). These results are shown in Table 7.

Table 6. Microbleeding by scraping method (n = 30).

*p < 0.05, McNemar’s test; **p < 0.01, Fisher’s exact test; In a comparison between TB3 and TB10, the number of participants with microbleeding was significantly higher in TB10. In a comparison between TongueB10 and TongueB30, the number of participants with microbleeding was significantly higher in TongueB30. A comparison of the 10-stroke groups showed that TB10 had a significantly greater number of participants with microbleeding than TongueB10 and S10. A comparison of the 30-stroke groups showed that TongueB30 had a significantly greater number of participants with microbleeding than S30.

Table 7.Factors associated with the absence of minor bleeding during tongue mucosa scraping (Generalized Linear Mixed Model).

OR > 1 indicates a higher likelihood of “no bleeding”; Interaction between device and stroke count was not significant (p = 0.126). Generalized Linear Mixed Model (Binomial distribution, Logit link function). Device and stroke count were included as fixed effects. Participant ID was included as a random effect. *p < 0.05.

3.3.1. Medical History and Medications

No significant associations between medical history or medication use with the number of participants with microbleeding were observed (Table 4, Table 5).

3.3.2. Nutritional Status

Assessments of nutritional status performed with the GNRI showed no significant differences in number of participants with microbleeding between the no/mild risk group and moderate/severe risk group (Table 8).

Table 8.GNRI (n = 27).

Fisher’s exact test. GNRI: Geriatric Nutritional Risk Index. No significant differences in number of participants with microbleeding were observed between the no/mild risk and moderate/severe risk groups for any of the scraping methods.

3.3.3. Tongue Coating

Tongue coating was evaluated using the Tongue Coating Index (TCI), and the median values in the S10 and S30 groups were significantly lower in participants with (8.33%) as compared to those without microbleeding (61.1%) (Table 9; Figure 1,Figure 2). There were no significant differences in TCI observed among the TB3, TB10, TongueB10, and TongueB30 groups.

3.3.4. Oral Moisture

When oral moisture was assessed using the clinical diagnostic criteria for oral dryness, the number of participants in the S10 and S30 groups with microbleeding by the scraping method differed significantly, with the moderate/severe dryness group showing a significantly higher incidence as compared to the normal/mild group (Table 10). In contrast, when assessing oral mucosal moisture levels using an oral moisture meter and number of microbleeds associated with each scraping method, no significant differences were observed (Table 11).

Table 9.TCI (n = 30).

t-test. **p < 0.01. In the S10 and S30 groups, TCI values were significantly lower in participants with microbleeding.

t-test, **p < 0.01. In the S10 group, TCI was significantly lower in participants with microbleeding.

Figure 1.Relationship between TCI and microbleeding in S10 group.

t-test, **p < 0.01. In the S30 group, TCI was significantly lower in participants with microbleeding.

Figure 2.Relationship between TCI and microbleeding in S30 group.

Table 10.Oral dryness (n = 30).

Fisher’s exact test. *p < 0.05, **p < 0.01. Clinical diagnostic criteria: Grade 0 (normal): No oral dryness or increased salivary viscosity. Grade 1 (mild): Increased salivary viscosity and slightly reduced salivary volume; saliva appears stringy. Grade 2 (moderate): Markedly reduced salivary volume; fine bubbles observed. Grade 3 (severe): No saliva visible on tongue mucosal surface. In the TongueB10, S10, and S30 groups, the number of participants with microbleeding was significantly higher in the moderate/severe group as compared with the normal/mild group.

Table 11.Oral mucosal moisture (n = 30).

Fisher’s exact test. Classification criteria: Normal: ≥29.6; Borderline: 28.0 - 29.5; Dry: ≤27.9; No significant differences in number of participants with microbleeding were observed among the scraping methods according to oral mucosal moisture level.

3.3.5. Number of Present Teeth

In the TB3 group, participants with fewer present teeth had significantly higher rates of microbleeding. Additionally, participants in the TongueB30 and S10 groups with fewer present teeth also showed significantly higher rates of microbleeding (Table 12). Five of the participants had 20 or more present teeth and none of those exhibited microbleeding under the TB3, TongueB10, TongueB30, S10, or S30 conditions.

Table 12.Present teeth (n = 30).

t-test. *p < 0.05, **p < 0.01. In the TB3, TongueB30, and S10 groups, participants with fewer present teeth showed a significantly higher incidence of microbleeding.